Session 1/15
Soil Properties of Fort Union Clay Shale
Propriété des argiles schisteuses de Fort Union
by C a r n e a l K . S m i t h , Head, Soils Section Garrison D istrict, M issouri River D ivision Corps o f Engineers, D epartm ent o f the
Army, Riverdale, N orth D akota, U .S .A ., and J. F. R e d l in g e r , Engineer, Soils Section, Garrison D istrict, M issouri River D i­
vision Corps o f Engineers, D epartm ent o f the Army, Riverdale, N orth D akota, U .S .A .
Summary
This paper gives the geological description and discusses the soil
classification, laboratory tests (including extra high load consoli­
dation test) on undisturbed samples and field data on the Fort Union
clay shale. At Garrison Dam, the largest rolled fill dam in the world,
the Fort Union clay shale is the foundation for heavy structures and
clay shales from the structural excavation are used in the rolled
embankment.
Introduction
Fort U nion clay shales are predominantly lean and fat clays
that have at one tim e been consolidated under loads estimated
at between 80 and 100 tons per square foot. They underlay
a region o f 12 0 ,0 0 0 square miles in area which includes the site
o f Garrison D am , at this tim e the largest rolled fill dam in the
world which is being designed and constructed by the Corps
o f Engineers, D epartm ent o f the Army. The F ort U nion clay
shales n ot only form the major portion o f the foundation for
heavy structures (intake, powerhouse and spillway) but the
clay shales from the structural excavations are used in the rolled
embankment. The size o f the em bankm ent (210 feet high,
11,500 feet long, containing 70 m illion cubic yards o f material)
and related structures make it im portant to have a thorough
knowledge o f the foundation material.
Geological Description
Garrison D am and Reservoir are located in the western half
o f the state o f N orth D ak ota which lies in the northern part
o f the Great Plains physiographic province known as the M is­
souri Plateau.
R oughly dividing this region into glaciated (east) and un­
glaciated (west) portions is the M issouri River which flows in
a southeasterly direction through the state in a series o f east-
west and north-south trending steps. G lacial history o f the
62
Sommaire
Ce rapport analyse la géologie du barrage de Fort Union et dis­
cute la classification du sol, les essais de laboratoire (y compris
l’essai de la charge maximum de consolidation sur des échantillons
non remaniés) et les observations sur place du schiste argileux de
Fort Union. Au barrage de Garrison, le plus grand barrage en terre
roulée du monde, le schiste argileux de Fort Union forme la fonda­
tion des constructions et les schistes argileux provenant de l’excava­
tion naturelle sont utilisés pour le massif roulé.
M issouri River is som ewhat com plex and continental glacia­
tion has greatly influenced its present channel. There is evidence
that the once eastward flowing streams with drainage into the
north were truncated by the M issouri which was itself forced
to take a southerly course after being blocked by ice on the
north and east.
The present stream flows in a valley which is 2 to 4 miles
wide and 200 to 400 feet deep. The valley itself (see Fig. 1)
has been cut m uch deeper than the level at which the present
river flows and in the past has also been filled with more
alluvium than at present. This is evidenced by borings in the
present flood plain which penetrate from 10 0 to 20 0 feet o f
alluvium and till before encountering bedrock and by terrace
remnants which may be seen at various levels above the river
throughout the valley.
The Fort U n ion group makes up the bedrock in a greater
portion o f the river valley within N orth D akota. The dam site
and probably the entire reservoir area lie in an upper portion
o f the group know n as the T ongue River form ation. This
group is o f early Tertiary (Paleocene) A ge, and except for the
lignite'beds, sands, and an occasional thin strata o f cemented
lim estone and sandstone, m ay be termed an immature (clay)
shale. The clay shales can be classified and tested by procedures
ordinarily applied to soils. Little cem enting material is present,
but the form ation has been consolidated by super-imposed
F ig. 1 Profile C enterline D a m . L o o k in g U p stream
C o u p e des lign es d e cen tre du barrage. D irectio n a m on t

loads o f similar material up to 1,500 feet in thickness. R e­
peated consolidation m ay have been brought about by at least
tw o ice sheets which probably covered the dam site area in
early Pleistocene times and terminated 50 miles more or less
west o f the river. Gradation o f the sediments which make up
the Fort U nion varies from fine sand to fat clay with lean clays
predom inating. H orizons o f an individual material are essen­
tially flat lying, often cross-bedded, and vary in thickness from
thin partings to over 15 feet, but the lateral extent o f any one
bed is limited. Prom inent throughout the form ation and es­
pecially at the dam site are beds o f lignite which are jointed
and cracked and are usually water bearing. T he thicker beds
may be traced for m any miles and are used as correlative
horizons. Slickensided fat clay layers are usually found adja­
cent to the lignite beds.
Soil Characteristics
Classification: F ort U n ion clay shales are predominately
lean and fat clays with som e sandy clays and clayey sands. The
results o f 175 Atterberg liquid and Plastic lim it tests are show n
on Fig. 2. A ll tests fall in the shaded zone above the A line.
In consistency they are hard and vary in color from light brown,
buff and grey in the weathered zone to more som bre and darker
colors in the unweathered zone. D r. Ralph E. Grim o f the
Illinois State G eological Survey reports the results o f a differ-
ential thermal analysis on a sam ple o f fat clay as follow s:
quartz (10 to 15 percent); clay minerals, 85 to 90 percent illite,
1 0 to 15 percent m ontm orillonite; 0 percent calcite and 1 per­
cent plus or minus organic matter.
D ensity and moisture content: In dry density, the Fort U nion
varies between 95 and 115 pounds per cubic foot, and its m ois­
ture content ranges from 16 to 24 percent.
Specific gravity: Specific gravity varies from 2.66 to 2.72.
Laboratory Tests
Shear tests: A summary o f direct shear tests is show n on
Fig. 3 for consolidated drained (slow) and consolidated un-
drained (consolidated quick) tests. In both types o f direct
shear tests the sam ples were consolidated under a load equal
or greater than their present overburden pressures and then
sheared. In the consolidated drained tests the horizontal load
was increased in increments and permitted to drain fully for
each increment. In the consolidated undrained, the sam ples
were quickly sheared under constant strain. Extreme variations
o f clay shale found in the form ation and in m any o f the indi­
vidual sam ples accounted for the large spread in the test re­
sults. A cohesion o f 0.7 ton per square foot and a coefficient

100

1 y
80
f,'" " ,

X
LJ
£
>
►
—
y
40
"a " l N E
co
<
\ »
^20
\ n
/ ! Ï B " l IN E
\S>>
f '
0 20 40 60 80 100 120 140 0 1 2 3 4 5 6 7 8 9 10
L IQ U ID L IM IT N O R M A L L O A D IN T O N S / S O F T '

Fig. 2 P lasticity C hart F ig. 3 R esults o f D irect Shear T ests

D ia g ra m m e de p la sticité R ésu ltats des essais de cisaillem en t direct

63
 0-500

0-400
o
h-
■St
o:
Q
5
>

0 300

0- 20 0
P R E S S U R E - TONS P E R SQ U. F T .

Fig. 5 T y p ica l R esu lt o f a L o a d in g T est

R ésu ltat typ ique d ’un essai de charge

from 800 x 10-4 to 200 x 10 -4 cm per second. This coefficient

was determined by pum ping tests. The clay shales have perme­
ability coefficients varying from .0 1 x 1 0 -4 to .0 0 0 1 x 1 0 -4 cm
per second.

Fig. 4 Sp ecial L o a d in g D ev ice F ie ld D a ta

D is p o sitif spécial de charge
Slope curves: Studies o f natural and excavated field slopes
o f friction o f tan. 2 0 degrees was used for the design o f ex­ in F ort U n ion were conducted to check analytical studies be­
cavated slopes. Zero cohesion and a coefficient o f friction o f fore designing the permanent excavated slopes for the Garrison
tan. 29 degrees was used for foundation design on unslick- Project. A field slope chart, Fig. 6 , summarizes the measured
ensided clays and tan. 25 degrees for slickensided clays. Pore slopes. The solid line curves o f Fig. 6 are the factors o f safety
pressures were calculated when using these latter values. from stability studies using a cohesion o f 0.7 tons per square
Consolidation: T o check the preconsolidation load a special fo o t and coefficient o f friction o f tan. 2 0 degrees and circular
loading device was constructed as show n by the photograph, arcs o f failure. It is seen that the curve for factor o f safety o f
Fig. 4. Loads o f 500 tons per square foot were obtained with 1 .0 goes very close to the point representing the highest and
this apparatus. A typical result is shown on Fig. 5. The results
show ed that the preconsolidation load could have been som e­ 200
where between 80 and 100 tons per square foot. Other tests
on more conventional apparatus were performed up to 50 tons
per square foot, and were cycled m any times. These tests re­ Ll ) 12
vealed that after the first cycle the com pression index and £ 150
swelling index did not vary appreciably. In its history the 210
form ation has been repeatedly loaded and unloaded. It is LU
CL
Y TEMPORARY
EX C A V A TED S L O P E S
know n from the elevation o f buttes in the surrounding region O
9 NATURAL S L O P E S
that at least 1,500 feet o f sediments once overlay the dam site w 100
u. A F T E R FA ILU RE
and from surface evidence that it has been subjected to at least o O N A TU RA L S LO P E S
tw o glaciations. A t the present tim e the form ation is on a re­ i □ P R O B A B L E SLO P E
bound branch o f one o f its m any cycles. After studying the o B EFO R E F A IL U R E
uj 50 A E X IS T IN G E X C A V A T E D
test results, it was decided to use a com pression and swelling
SLO P E S
index o f 0.018 for m ovem ent com putations. The coefficient o f
F.S.FACTOR OF S A F E T Y
consolidation when plotted against the average, pressure on
COHESION 0 - 7 0 T / S 0 FT
sem i-log paper is a straight line passing through 2 x 10 ~ 4 cm 2 20
per second at an average pressure o f 4 tons per square foot 0 0 5 10
and 9.5 x 10 -4 cm 2 per second at 20 tons per square foot. COTANG E N T OF SLOPE
Perm eability: The m ost permeable member o f the Fort Fig. 6 F ield S lo p e C hart
U nion is the lignite which has a permeability coefficient varying D iagram m e des pentes des talu s naturels

64
steepest slope as well as the point for one o f the slopes that
failed. This study o f actual slopes in Fort U nion is considered
a check on the adopted shear strength values for large masses
o f Fort U nion. In the final design o f the permanent slopes for
Garrison Project, the effects o f lowering the ground water and
the variation o f reservoir p ool were considered.
E lasticity: Fort U n ion clay shales are elastic and large rapid
settlem ent and rebound occurs on loading and unloading. Both
a time settlem ent due to consolidation and rapid settlement
due to elasticity were calculated for all structures. By perform­
ing triaxial tests called constant stress ratio tests (the lateral
load is kept at a constant ratio to the vertical load as the verti­
cal load is increased) the m odulus o f deform ation for a stress
ratio o f one-half was determined as 2 ,0 0 0 tons per square foot.
Subsequent measured rebounds and settlements are approxi­
mately equal to the predicted value based on this constant.
L ateral Pressures: W hen saw cuts 3 inches wide and 7 feet
deep were made at Elev. 1612 in the powerhouse area, the cuts
closed in about 24 hours. This was attributed to yield caused
by the release o f trem endous lateral pressures that were induced
by the overburden, which in the past ages was more than 1,500
feet higher than at present. The elevation o f these cuts is about
the minim um average elevation that the M issouri had ever cut
in the Fort U nion at the dam site as shown in Fig. 1. These
Fig. 7 Studies o f th e E ffect o f U n lo a d in g in the E ast A butm ent
E tudes des résultats de d écharge sur l’appui latéral est
high lateral pressures once may have existed in the formation
at higher elevations, but the valley cut by the M issouri in the
Fort U nion allowed this higher material to yield laterally,
thereby releasing this pressure. The floor o f the valley pre­
vented the lateral yield o f layers at Elev. 1615 and lower.
C h aracteristics B ased on F ield T ests
Pressure Void Ratio Curves based on Field D ata: The bluffs of
the east abutm ent o f Garrison D am and the east flood plain
differ in elevation by more than 20 0 feet and furnish a means
o f estim ating the effect o f unloading by erosion on the under­
lying strata o f a large mass o f material. A study was started
in 1948 to estimate by field observations what effect this un­
loading caused and to see if it checked predictions based on
laboratory data. In the bluff, H ole 972, surface elevation o f
1924, was drilled to Elev. 1574 and in the east flood plain H ole
971, surface elevation 1693 was drilled to an elevation o f 1574.
A profile showing the relative positions o f the borings is shown
in Fig. l a . In both borings below Elev. 1660 undisturbed soil
sam ples were taken at every change o f material. Liquid, plastic
limits, moisture and density tests were performed on the samples
and the results are show n in Fig. l b . From the densities and
moisture contents the void ratios were calculated for each
65
sample. Pressure-void-ratio curves were drawn, as shown in
Figs. 7 c and d, for pairs o f samples. The basis o f selecting a
pair o f samples was that a sam ple from H ole 972 was a mate
o f a sam ple from H ole 971 if it had approximately the same
liquid and plastic limits and their elevations varied similar to
adjacent lignites. It was assumed that the existing pressure for
any sam ple was equal to the present weight o f material over
the sample. The watertable was assumed at about Elev. 1690.
The average slope o f the pressure-void-ratio curves derived
from the above field study do not check the value o f the ex­
pansion index o f 0.018 adopted from laboratory tests. The
average value from field tests was 0.05. It is believed that the
reason for this difference is that the field data includes the long
time secondary com pression and swell (creep) which was not
included in the laboratory data. Instead o f a line being drawn
directly from void ratio o f a sam ple in H ole 972 to that o f its
mate in H ole 971, the soil could have follow ed the probable
curve solid line from points 13 to 51 as show n in Fig. 7 c. This
w ould be the probable curve if the unloading had occurred
relatively fast and then conditions remained the sam e over a
long period o f time. Similar conclusions have been arrived at
by A. Casagrande (1949) in his studies o f Saskatchewan Bear-
paw Shale.
Check o f characteristics using rebound o f lignite beds: Logs
o f H oles 971 and 972, show n on Fig. l b , show that the lignite
beds o f the flood plain (H ole 971) are a distance o f 7.4 to 4.6
feet above similar beds in H ole 972. The lignite beds were
probably level at the tim e they were laid down, and if the
material below H oles 971 and 972 was uniformly compressible,
the elevations o f the lignite layers would furnish a measure o f
rebound caused by erosion o f the valley. The rebound o f 1L
lignite bed was calculated, using the previously m entioned
m odulus o f deform ation o f 2 ,0 0 0 tons per square foot and
expansion and com pression index o f 0.018. N o consideration
was given to the secondary effects in the calculations. It is im ­
probable that the form ation in H oles 971 and 972 w ould have
the sam e compressibility layer for layer, as the holes are 653
feet apart, but the depth o f form ation effecting the rebound,
600 feet, is so great that it was assumed that the average charac­
teristics o f the form ation affected were the same.
F or purposes o f calculation the original surface o f the ground
66
and water table was assumed to be the top o f H ole 972. A s the
water lowered and the river cut deeper and deeper into the for­
m ation, the lignite beds were exposed in the faces o f the river
bluffs, allow ing the water in the beds to drain out. Ground
water in the vicinity o f the bluffs tended to seep vertically to
the lignite beds, which act as horizontal drains. These two
actions, the erosion o f the valley which caused the form ation
below the valley to rebound and the lowering o f the ground
water table which caused settlem ent o f the form ation below
the bluffs, com bine to create a maximum difference in elevation
o f the 1L lignite between H oles 971 and 972.
Calculations summarized in Fig. l e show that the rebound
was 2 .6 feet, 1 .0 fo o t due to swelling and 1 .6 feet due to elastic
rebound, while the settlem ent caused by the lowering o f the
water table in the bluffs was 0.5 fo o t consolidation and 1.1 feet
elastic com pression. The sum o f these effects is 4.2 feet. The
actual measured difference in elevation o f 1 L lignite between
H oles 971 and 972 is 7.4 feet and that for the 2L is 5.6 feet.
T he difference between the measured values and the calculated
values could be due to the com bined secondary compression
(creep) o f the soils represented by H ole 972 reducing in v o ­
lum e and those o f H ole 971 increasing in volum e.
C o n c lu sio n s
A t this time certain significant conclusions seem to stand out
concerning the Fort U nion clay shales. O f major im portance
is that the unloading o f Fort U nion clays even for lon g periods
o f tim e does not reduce its strength enough to endanger large
structures; secondary com pression and expansion can be o f
the sam e m agnitude as consolidation; large horizontal forces
exist in the clays below the elevations o f eroded valleys in the
form ation; the C L clays usually swell more than the C H clays.
(The cause o f this is not understood at this time.)
A com panion paper to this, “ R ebound G ages Check M ove­
m ent Analysis at Garrison D a m ” , com pares the field data with
the calculated rebound and settlement.
Reference
C asagrande, A . (1949): N o te s o n Sw ellin g C haracteristics o f C lay-Shales.
H arvard U n iversity, P ierce H a ll, C am brid ge, M ass., July, p. 12.